Support device and method

1 . An assistance device for assisting a determination inspection for detecting, from a captured image of an object to be inspected, an abnormality occurring in the object to be inspected and determining a type of the detected abnormality, the assistance device comprising:

a classification processing unit; and

a learning discrimination unit,

wherein the classification processing unit is configured to:

calculate, from the captured image, a feature amount of the abnormality occurring in the object to be inspected:

classify the calculated feature amount into a plurality of classes; and

select, from the plurality of classes, one or more classes to be processed, for each of which an abnormality type is to be discriminated using the learning discrimination unit,

wherein the learning discrimination unit includes a learning discriminator for each abnormality type, the learning discriminator being trained with a discrimination criterion by using set data obtained by combining an element image causing the abnormality and an abnormality type corresponding to the element image as a correct answer, the discrimination criterion being for discriminating whether an abnormality to be discriminated is of the abnormality type being a correct answer or of an abnormality type other than the abnormality type being a correct answer,

wherein the learning discrimination unit discriminates an abnormality type for each of the one or more classes to be processed, selected by the classification processing unit, based on an individual discrimination result from the learning discriminator trained with the discrimination criterion for each abnormality type, and

wherein the learning discriminator includes a first learning discriminator trained with a first discrimination criterion for discriminating a first abnormality and a second learning discriminator trained with a second discrimination criterion for discriminating a second abnormality, and when, for one class of the plurality of classes to be processed, a discrimination result from the first learning discriminator identifies the first abnormality and a discrimination result from the second learning discriminator identifies an abnormality type other than the second abnormality, the abnormality type for the one class is determined to be the first abnormality.

2 . The assistance device according to claim 1 , wherein when, for one class of the plurality of classes to be processed, a discrimination result from the first learning discriminator identifies an abnormality type other than the first abnormality and a discrimination result from the second learning discriminator identifies the second abnormality, the abnormality type for the one class is determined to be the second abnormality.

3 . The assistance device according to claim 1 , wherein when, for one class of the plurality of classes to be processed, a discrimination result from the first learning discriminator identifies the first abnormality and a discrimination result from the second learning discriminator identifies the second abnormality, or a discrimination result from the first learning discriminator identifies an abnormality type other than the first abnormality and a discrimination result from the second learning discriminator identifies an abnormality type other than the second abnormality, discrimination of an abnormality type for the one class is handed over to specified processing.

4 . The assistance device according to claim 1 , wherein the classification processing unit further classifies into a plurality of subclasses a feature amount of a class, for which the abnormality type has been determined.

5 . A method performed by a computer of an assistance device that assists a determination inspection for detecting, from a captured image of an object to be inspected, an abnormality occurring in the object to be inspected and determining a type of the detected abnormality, the method comprising:

calculating, from the captured image, a feature amount of the abnormality occurring in the object to be inspected:

classifying the calculated feature amount into a plurality of classes;

selecting a predetermined class from among the plurality of classes;

learning, for each type of the abnormality, a discrimination criterion by using set data obtained by combining an element image causing the abnormality and an abnormality type corresponding to the element image as a correct answer, the discrimination criterion being for discriminating whether an abnormality to be discriminated is of the abnormality type being a correct answer or of an abnormality type other than the abnormality type being a correct answer; and

discriminating an abnormality type for the predetermined class, based on an individual discrimination result obtained by discrimination using the discrimination criterion learned for each type of the abnormality,

wherein learning the discrimination criterion includes learning a first discrimination criterion for discriminating a first abnormality and a second discrimination criterion for discriminating a second abnormality, and when, for one class of the plurality of classes to be processed, a discrimination result from the first discrimination criterion identifies the first abnormality and a discrimination result from the second discrimination criterion identifies an abnormality type other than the second abnormality, the abnormality type for the one class is determined to be the first abnormality.